All statements of fact, opinion, or analysisexpressed in this monograph are those of theroundtable participants. They do not necessarilyreflect official positions or views of the CentralIntelligence Agency or any other US Governmententity, past or present. Nothing in the contentsshould be construed as asserting or implying USGovernment endorsement of the study’s factualstatements and interpretations.

Central Intelligence Agency

Center for the Study of Intelligence

Center for the Study of IntelligenceRoundtable Report

Intelligence and Policy:

The Evolving Relationship10 November 2003

Georgetown University,Washington, D.C.

Washington, D.C.June 2004

iii

Introduction

As became abundantly clear during a conference sponsored by CIA's Center for the Studyof Intelligence (CSI) in Charlottesville, Virginia, on 10 and 11 September 2003, thechallenges that face the US Intelligence Community in the aftermath of the terrorist attackon the United States two years earlier are perceived by members of that community asbeing far more complex, demanding, and consequential than any they have heretoforeencountered. That conference brought together an experienced group of national securityspecialists from the intelligence and policy communities to discuss Intelligence for a NewEra in American Foreign Policy.

Not long after the Charlottesville conference, Dr. James Steiner, CIA’s Officer inResidence and Associate at the Institute for the Study of Diplomacy (ISD) of the EdmundA. Walsh School of Foreign Service at Georgetown University, coordinated an effort toanswer one of the challenging questions that have arisen in the changed post-9/11 securityenvironment: how can the Intelligence Community effectively provide "actionable"intelligence while being mindful of its traditional practice of separating, to the extentpossible, the intelligence and policy functions of national security decisionmaking. Theresulting one-day roundtable conference became for CSI the first in a planned series ofprojects on intelligence and policy intended to foster better understanding of the often-perplexing dynamic between the consumers of intelligence and intelligence professionals.

The roundtable, Where Is the Red Line? Actionable Intelligence vs. Policy Advocacy,took place on 10 November 2003 at Georgetown University. Instead of using a conferenceformat, with formal papers and designated commentators, the roundtable was conductedas a discussion among a relatively small circle of participants, divided about equallybetween professional (current or former) intelligence officers and senior intelligenceconsumers drawn from the ranks of former policymakers. Ambassador Thomas Pickering,former Under Secretary of State for Political Affairs, and a former Deputy Director ofCentral Intelligence, Richard Kerr, served as cochairmen.

In addition to the cochairmen, participants included:

• Frans Bax, President, CIA University

• Hans Binnendijk, Director, Center for Technology and National Security Policy, NationalDefense University; Special Assistant to the President and Senior Director for DefensePolicy and Arms Control, National Security Council, 1999–2001

• Dennis Blair, Admiral, USN (ret.); President, Institute for Defense Analyses; formerCommander-in-Chief, Pacific Command; former Assistant Director of CentralIntelligence for Military Support

iv

• Christopher Bolan, Colonel, US Army; ISD Associate; former member of the staffs of vicepresidents Gore and Cheney, focusing on Middle East issues

• Chester Crocker, James R. Schlesinger Professor of Strategic Studies, Edmund A.Walsh School of Foreign Service, Georgetown University; Chairman, US Institute ofPeace; Assistant Secretary of State for African Affairs, 1981–89

• James Dobbins, Director, International Security and Defense Policy Center, RANDCorporation; served in a variety of State Department and White House posts, includingAssistant Secretary of State for European Affairs and Ambassador to the EuropeanUnion; also served as US special envoy for Afghanistan, for Kosovo, for Bosnia, for Haiti,and for Somalia

• Carl Ford, Assistant Secretary of State for Intelligence and Research, 2001–2003;Principal Deputy Assistant Secretary of Defense for International Security Affairs,1989–93

• Paul Johnson, Director, Center for the Study of Intelligence

• Woodrow Kuhns, Deputy Director, Center for the Study of Intelligence

• Douglas MacEachin, staff member, National Commission on Terrorist Attacks Upon theUnited States (9/11 Commission); Deputy Director for Intelligence, CIA, 1993–95; SeniorResearch Fellow, John F. Kennedy School of Government, Harvard University,1995–2000

• John MacGaffin, former Senior Adviser to the Director and Deputy Director, FederalBureau of Investigation; former Associate Deputy Director for Operations, CIA

• William Nolte, Deputy Assistant Director of Central Intelligence for Analysis andProduction

• Phyllis Oakley, Chair of the Board, US Committee for the United Nations PopulationFund; Assistant Secretary of State for Intelligence and Research, 1997–98; andAssistant Secretary of State for Population, Refugees, and Migration, 1994–97

• Martin Petersen, Deputy Executive Director, CIA

• Jennifer Sims, Visiting Professor, Edmund A. Walsh School of Foreign Service,Georgetown University; Deputy Assistant Secretary of State for IntelligenceCoordination, 1994–98; Intelligence Adviser to the Under Secretary of State forManagement and Coordinator for Intelligence Resources and Planning, 1998–2001

• James Steiner, CIA Officer-in-Residence, ISD

v

• Casimir Yost, Marshall Coyne Professor in the Practice of Diplomacy, Edmund A. WalshSchool of Foreign Service, Georgetown University; Director, ISD

The following summary of roundtable proceedings does not attempt to recapitulate thediscussions in detail. It attempts, rather, to focus on the most salient points made by theparticipants as they considered a set of key questions drawn up in advance by theroundtable sponsors. These questions will be found at the conclusion of the text. Readerswill note that some of the questions were discussed more extensively than others.

Those interested in sampling the tenor of the discussions may refer to the italicizedexcerpts contained in each section.

vii

Page

vii

ContentsThe Policy Community-Intelligence Community Nexus 1

The Intelligence Community Views Its Customers 1

Discussion Excerpts 1

What Policymakers Want From Intelligence 2

Discussion Excerpts 3

The Evolving Role of the Intelligence Community 7

Keeping Up with the Competition 7

Discussion Excerpts 7

Challenges for Analysts 8

Discussion Excerpts 9

Challenges for the DCI 10

Discussion Excerpts 11

The Elusive Red Line 13

Discussion Excerpts 13

Key Questions 17

1

The Policy Community-IntelligenceCommunity Nexus

The Intelligence Community Views ItsCustomers

The observation of a former senior intelligenceofficer that, in his experience, intelligenceanalysts often knew more about the countriesthey followed than they did about thecustomers they served led roundtableparticipants initially to debate the nature of thetarget audience for Intelligence Communityproducts. One rather expansive definition heldthat anyone on the receiving end of anintelligence product could conceivably makepolicy, including, for example, a member of thearmed forces in the field who chose to takeaction on the basis of a tactical intelligencereport. Such a recipient, it was suggested,might, however, more aptly be considered apolicy “implementer” than a policy "maker."

Pursuing this line, a roundtable participantthought that "decisionmaker" might be a moreuseful definition in that a customer could wellbe a "policymaker" at one point and a “policyimplementer" at another point. Moreover, thespeaker suggested, as an intelligenceconsumer moved along this spectrum, thenature of his dialogue with analysts wouldchange, as would the products they providedin response.

Other participants preferred a more restrictivedefinition that excluded tactical-levelconsumers and focused on consumers at thepolicy level. These consumers would certainlyinclude the president, the cabinet, the cabinetdeputies, and those holding assistantsecretary-level positions in the variousdepartments. Speakers then suggested thatkey consumers might also include officialsgiven special, high-level assignments; keycabinet and congressional staff members; andthose heading delegations to importantnegotiations.

A former senior intelligence officer cautionedagainst trying to arrive at too precise adefinition of a policymaker, arguing thatidentifying the audience and matching product

and audience are part of the intelligenceprofessional's job. Further to that observation,a speaker noted that the number of officialswho see themselves as having a role to play inthe policy process has increased, as has thenumber of agencies they represent, both ofwhich increase the demands levied on theresources available to the IntelligenceCommunity.

Discussion Excerpts

By intelligence, you could mean one of twothings: you could mean information obtainedclandestinely, or you could mean any productof the Intelligence Community. I assume youmean the latter, since a lot of the IntelligenceCommunity products are derived from overtlyobtained material.

* * *

Some intelligence providers are also players—verification has been one, covert action isanother. Any intelligence operation overseashas some policy significance.

* * *

When you add covert action to the mix and tryto figure out who is the policy implementer ornot . . . is the CIA officer in the field withMasood an implementer or a collector? Theanswer is: "Yes."

* * *

This notion that policymakers can dointelligence as well as the IntelligenceCommunity is flat bullshit. There's too muchinformation. The volume is so great that anypolicymaker who believes that he can look atthat and come up with good answers is a fool.And you ought not to provide them withintelligence anyway. But the problem is thatthe Intelligence Community hasn't recognizedthat as well. Shame on us if we can't do better

In a sense, youcould start off bymaking theassumption thatit is inevitablethat, if themachinery workswell, intelligenceshapes policy.

. . . I would bemuch happierwith competentpolicymakerswho know a lotabout the subjectthey’re dealingwith, but knowenough to knowthat they won’tever know asmuch as a reallygood intelligenceanalyst.

2

than policymakers. We can get moreinformation out of what we're collecting; we'rejust not doing it.

* * *

In the last two administrations I had experiencewith, we spent a lot of our time trying to figureout how the people coming in functioned—what their biases and their interests were. Imean, we spent a lot of time on intelligencefocused on the principal players, and it wasworth every minute of it.

What Policymakers Want From Intelligence

Roundtable participants recognized thatpolicymakers desire both substantive andbureaucratic support from the IntelligenceCommunity. On the substantive side, they wantreliable information on new developments andon matters with which they are unfamiliar. Theyalso want intelligence to inform theirdecisionmaking by describing the choicesavailable to an adversary or an opposingnegotiator and explaining how and why onechoice or another might be preferred. On thebureaucratic side, they want intelligence togive them an edge in policy deliberations.Several speakers spoke admiringly of seniorpolicymakers who developed a closerelationship with their opposite numbers inintelligence in order to give themselves anadvantage over bureaucratic rivals.

Participants with an intelligence backgroundobserved that the policymakers they haveserved have had quite different approaches tothe Intelligence Community and different stylesin dealing with their analytical interlocutors. Forexample, some have begun by professing littleuse for intelligence and much confidence intheir own knowledge and ability to make policydecisions. Others have appeared awestruck bythe intelligence products they were offered.

These contrasting attitudes, it was noted, havegenerally reached a "crossover" point at whicha rough balance in approaches was achieved.

A speaker with experience in both policy andintelligence positions commented that mostpolicymakers failed to make efficient use of thecapabilities of the Intelligence Community,relying on analysts to think up the questionsthey should want answered. Other participantsadded a caveat, however. In their view, evenwhen policymakers actively solicit input fromthe Intelligence Community, intelligenceofficers must be wary of responding to thequestions in their own terms. Theseparticipants argued that, if necessary, analystsshould recast the questions to make sure thattheir analyses are not compromised by apartisan agenda and that the issues thatshould be addressed are addressed.

Several participants raised the question of biasand its role in causing intelligence failures. Thediscussion focused on several aspects. On theone hand, intelligence producers can beresponsible for failures through erroneousassumptions or personal prejudice.Consumers, on the other hand, often causefailures through reluctance to acceptintelligence they don't want to hear.

Continuing on this theme, participantsrecognized the practically limitless volume ofinformation, both classified and unclassified,that is now available to policymakers. This,they added, has led many policymakers toconclude that they can do their own analysis.Several speakers noted that, although manypolicymakers could point to some prior foreignpolicy experience, this approach could beharmful because it prevents the policymakerfrom taking advantage of the knowledge thatyears of study affords analysts.

There was agreement among participants that,despite its expense and inefficiency, havingmore than one intelligence agency competingfor the consumer's attention has generally

We can get moreinformationoutofwhat we'recollecting; we'rejust not doing it.

3

served the country well. Several speakersadded that they would not be troubled byadditional competition—one example beingboutiques created for a specific purpose byindividual policymakers—so long as allproducers were subject to the same rules. Aformer policymaker pointed out that the USmodel of an intelligence system differed fromthose of several key allies, which are eithermore restrictive or more freewheeling.

The same former policymaker remindedparticipants of the commonly held view thatintelligence analysts almost always tendtoward pessimism in their appraisals becausethey are more likely to be criticized for failing topredict an untoward event than for making acall that turns out to be wrong. Policymakers,on the other hand, tend to be optimists.Occasionally, however, when they are reluctantto take action on some issue, policymakers arehappy to receive a pessimistic assessment.Such conjunctures, he noted, work againstpolicy change and discourage innovativethinking.

A speaker commented on the difficultyanalysts encounter in gaining acceptance forscenarios with non-linear outcomes. Thisprompted another participant to lament thatwar gaming, which sometimes produced suchresults by bringing analysts and policymakerstogether in structured exercises, had fallen intodisuse in recent years.

Discussion Excerpts

But one of the things I observed is that someadministrations came in saying, “Intelligencecan't help me at all. I don't like it. I don't trust it.I am not confident in it. I have my own way ofthinking about problems, thank you very much.We'll take your stuff, but don't expect a greatdeal of interaction.” Another group came insaying, “This is an omnipotent group…[it]knows everything. I can hardly wait toembrace 'em.” All of them changed their views.

The ones that were skeptical when they camein became increasingly dependent or, at least,reliant on it. The ones that loved it at thebeginning began to say, “Is that all you can dofor me?” So you cross somewhere in the center. . . .

* * *

The goals of a player in a bureaucratic warfaregame are, first, control. Second, power. Andpower comes from expertise; so you needintelligence to win battles that are aboutexpertise, and you need intelligence to be ableto be effective with the foreigner as well.Thirdly, to achieve your goals, whatever thosegoals might be, to exploit those opportunities,to carry out that policy. And fourth on my list isto support the national interest. These are indescending order.

* * *

In the policy formulation process, people canact very tactically as they're trying to grab arole in doing policy. Policymakers, on any givenissue, aren't necessarily a static set. They arecompeting to get into the policy domain, andthey want intelligence to support them in thetactical process of getting an “in” on an issue.

* * *

Quite often, policymakers do not wantintelligence on a problem they do not want tohear; or they've already heard the answer, andthey don't like it.

* * *

Policymakers, fundamentally, are doers,particularly early in an administration. Theytend to have strong personalities. They comein with an agenda, and, often, they think theycome in with a mandate. And, therefore, I think

. . . power comesfrom expertise;so you needintelligence towin battles thatare aboutexpertise . . .

4

what they are looking for from intelligence,particularly early on, is information that helpsthem push that agenda.

* * *

I think the intelligence suppliers have tounderstand that policymakers have an agenda.They have both a bureaucratic one, and theyhave, perhaps, a substantive agenda.Bureaucratically, they're trying to enhance theirown position in a kind of zero-sum game vis-à-vis everyone else in the same sphere.Substantively, they probably have a set ofobjectives they're committed to trying toachieve. And that means that they will respondto some kinds of intelligence differently thanothers—that they will respond better tointelligence that enhances either their positionor their chances of achieving their substantiveobjective and that they will at least regard moreskeptically intelligence which has the oppositeeffect.

* * *

Probably you'll go back and find mostintelligence failures are not because of lack ofinformation, but because of assumptions andpredictions that were based on biases.

* * *

A colleague once used weather forecasting asopposed to predicting: “I've got high pressure,so much temperature. If nothing changes, andif my assumption about this is correct, this iswhere the thing's going to come out.” But, if I sitdown and say, “This is how they think, this iswhat they'll do, and, therefore, this is theoutcome”—that's the recipe for disaster.

* * *

But the terms of reference [of a NationalIntelligence Estimate] were defined not by theAgency, but by the requester . . . in my view avery dangerous process, and a thing that had

been resisted, at least in my experience, for aslong as I could remember. And that is, whenthe Hill asked for an estimate, or thepolicymaker asked for an estimate, that's fine.But recast the terms of the estimate in termsthat you wish to address.

* * *

When someone sits at a desk at the NSC andsees all of the raw take, does that mean thatthat policymaker doesn't have to rely on theIntelligence Community as much as he or shewould have a decade ago or two decades ago,and does that make it easier for thepolicymaker to say, “I don't need the intelcommunity; I'm going to have my own analystssitting here taking a look at this and reachingour own conclusions, and, therefore, I can cutout the intel community?”

* * *

If somebody wants to create his own littleintelligence cell, fine. But, subject it to thesame competition that the rest of us have to gothrough. But what the problem is,organizations are created. They're notsubjected to the kinds of competition andtransparency [as the rest of us].

* * *

By and large, my experience is that almost allthe analytical agencies have the same stuff,and, so, then I think what good decisionmakersvalue is our different opinions. I mean, the bestpart of NIEs used to be the footnotes.

* * *

There's an interesting distinction between theAmerican intelligence system and the Britishintelligence system . . . and the German. TheAmerican, in principle, is open and competitive.That is, everybody has access to all theinformation. Each of the agencies can come toits own conclusions. They are not forced to a

If somebodywants to createhis own littleintelligence cell,fine. But, subjectit to the samecompetition thatthe rest of ushave to gothrough.

5

common conclusion, and they can brief thoseconclusions to anybody they want. The Britishsystem is open, but coordinated. That is,intelligence agencies aren't permitted to brieftheir conclusions to policymakers exceptthrough the Joint Intelligence Committee,which is chaired by a policymaker and whichcomes to a coordinated judgment, and only thatcoordinated judgment goes to thepolicymakers. And [there's] the Germansystem, which is neither transparent norcoordinated; in which the intelligence agenciessecretly pass such information as they chooseto such policymakers as they wish.

* * *

But one of the things that, it seemed to me,rescued us a number of times was a deep,deep understanding of the facts and the issues

at a level of detail where you could go into aroom and defend your argument. I do have theimpression right now that we go into roomsunarmed to deal with serious critics who knowa lot.

* * *

I think policymakers and decisionmakers canfancy that they are, in fact, better than theintelligence analysts because they get thesame data. The only thing they don’t have is 30years’ worth of looking at this country and whatthat does in the brain cells in the back of thebrain, which it doesn’t if you’ve been doingother things for 30 years.

But one of thethings that . . .rescued us anumber of timeswas a deep, deepunderstanding ofthe facts and theissues at a levelof detail whereyou could go intoa room anddefend yourargument.

7

The Evolving Role of theIntelligence Community

Keeping Up with the Competition

Participants recognized that there has been amarked change in the role of intelligence sincethe end of the Cold War. Until the 1990s, theIntelligence Community virtually "owned"information on the USSR and the communistworld, the principal strategic challenge facingthe United States and its allies, because mostof that information was acquired clandestinelyor technically. Developments of the past 15years have vastly increased the amount ofinformation available to policymakers,however, and have deprived the IntelligenceCommunity of its dominant position. A speakeradded that the volume of available informationis growing steadily as a result of the currentemphasis on collection.

Several participants saw problems in theincreasing focus of intelligence producers oncurrent intelligence and policy support at theexpense of basic research. In their view, whilethe Intelligence Community does a good job ofproviding policymakers with currentintelligence, this shift in emphasis hasproduced a lack of analytic depth that all toooften makes intelligence products little or nobetter than what most reasonablysophisticated policymakers can provide forthemselves. The problem is made more acutein that a large percentage of the intelligenceworkforce is relatively new on the job.

Other speakers, arguing that policymakersperforce focus on events of the moment andare able to peruse only a relatively smallamount of the information available to them,advocated efforts by analysts to takeadvantage of their ability to concentrate on aricher store of information to look for ways togive policymakers products that provideneeded context for the intelligence reports thatcross their desks. Carrying the argument astep further, another speaker commented onthe importance of analysts' having sufficientknowledge of their fields, as well as anunderstanding of the way in which policies are

developed, so as to be able to givepolicymakers intelligence analyses they maynot even realize they need.

Another participant, a former high-levelintelligence officer, contended that theIntelligence Community's efforts to maintainanalytical relevance should include arequirement for systematic critiques ofCommunity products. In addition, the inclusionof policymakers in such critiques would make itmore likely that the needs of the policycommunity were taken into account. Thisobservation provided a counterpoint to anearlier suggestion that, to the extent possible,policymakers should include intelligenceofficers in their meetings, so that policycommunity concerns and needs might beconveyed to their colleagues more accuratelyand expeditiously.

Discussion Excerpts

I think there is a fundamental change in therole of intelligence since the end of the ColdWar, where intelligence provided the bulk ofthe knowledge and information on the strategicproblem. And it was secret. It wasclandestinely acquired, or technically acquired,and, therefore, intelligence essentially ownedthat information. Today, there is no ownershipof information by the Intelligence Community.

* * *

The Intelligence Community really [is] focusedon current intelligence, on policy support. Itdoes very little research. It has very littleunderstanding below the level of thepolicymaker and, in my view, on many issues.I think that, in some ways, these two groupsare reinforcing each other's worst habits.

* * *

Today, there is noownership ofinformation bythe IntelligenceCommunity.

[Intelligence] hasvery littleunderstandingbelow the level ofthe policymakerand, in my view,on many issues. Ithink that, insome ways,[intelligence andpolicymakers]are reinforcingeach other'sworst habits.

8

If you focus on current intelligence, that's aboutten percent of the information available to theIntelligence Community. The IntelligenceCommunity is really the only one in town thathas the time to look at the other 90 percent andfind the things that don't stick out to the currentintelligence officer or the policymaker the firsttime around. In fact, my observation is thatpolicymakers often know more than theintelligence officers, particularly the seniorones, because they've been on the telephoneto the King of Tut or to the president of Wa. Andthey talk to people all the time. They talk to thePresident; they know what the Secretary ofState and the Secretary of Defense know, andintelligence officers don't have any real accessto that sort of knowledge. The fact is that mostpolicymakers are starving for new knowledge.And if it's good, if it's new, I don't care howskeptical they are, you can sell it to them.

* * *

Your average analyst today, and certainly oneworking terrorism, is probably seeing in thetenths of what's available. Furthermore, we'repouring so much money into collectionsystems and very little into the exploitation ofthat collection, that the problem is gettingbigger. We've got to put more money into theanalytic side and balance the collection, orwe're just going to be another opinion.

* * *

The sheer volume of information, the sheergrowth of consumers, the pressure to do itquickly, has driven research out of the market.

Challenges for Analysts

The discussion of how the analyticalcommunity should most effectively packageand deliver its product to the policy communityled to an exchange on the content of thoseproducts. There was ready agreement that oneof the major developments since the terrorist

attacks of September 2001 is that the standardfor analytic success has changed dramaticallyfrom that of the Cold War, when the question ofwhether or not intelligence was performing wellagainst the Soviet target lacked practicalrelevance for most Americans. Now, on theother hand, in the war against terrorism, publicexpectations of intelligence have becomeunreasonably high—as one speaker put it, "likeexpecting the FBI to stop bank robberiesbefore they occur."

Not surprisingly, there was also generalagreement that analysts must strive to avoidboth strict reportage and outright advocacy ofpersonal points-of-view. Drawing on hisextensive experience supporting USdelegations, one speaker argued that analyticadvocacy had to be distinguished from analyticadvice, in which the service that intelligenceperforms is to describe alternatives and theirpotential consequences, including those thatpoint out flaws in positions policymakersfavored and those that present the least bad ofa series of bad choices. Another speakercautioned that analysts should not dwell onpersonal perceptions of the correctness orincorrectness of policy decisions, noting thatsuch fixations occasionally cause higher-upsto "jerk" the system in order to redirect its focustoward the current situation and its actualalternatives. (He added that some might findthat such a directed shift in emphasisconstitutes "politicization.")

Further to this discussion, participantscommented on the contrast between militarypractice—where subordinates are expected toargue their points-of-view vigorously until adecision is made, after which they areexpected to give full support to that decision—and that of intelligence, where the fact that apolicy decision has been made does notrequire analysts to cut their analyses to thatcloth. It was noted that their insistence on"speaking truth to power”—mentioned by anumber of observers as a long-timeDirectorate of Intelligence cultural trait—has

The sheer volumeof information,the sheer growthof consumers,the pressure todo it quickly, hasdriven researchout of the market.

9

frequently caused analysts to be regarded asless than welcome guests at the tables wherepolicy is debated.

Several participants wondered if this culture ofspeaking forthrightly extended to internalAgency deliberations on such matters, forexample, as covert action. In response, severalspeakers argued that an independent stancewas not only possible but a necessary aspectof analytic "checks and balances," even if thecritiquing process led to in-house clashes. Aformer senior intelligence officer remindedparticipants that the failed Bay of Pigsoperation of 1961 had been one from which theanalytic side of the house had been completelyexcluded.

Discussion Excerpts

A former senior intelligence officer: It's seldomthat the Intelligence [Community] looks forpolicy opportunity. I mean, looks at the goodside. I mean, it doesn't say, "Boy, here'ssomething that's really interesting that youcould do."

A former senior policymaker: Well, that's notyour day or night job yet. It's somebody else's.

* * *

I think where we want to be is in that middleground between the Sherman Kent school,which basically dominated the estimativeprocess up until 1973, and where we've comesince then. Sherman Kent's view was that itwas important for the estimative process prettymuch to be detached from policy and to behighly objective. So, [by] making sure that theintelligence community had an ivory tower todo that basic research, you would be providingsomething the policymaker couldn't get in hisbusy day. But the downfall of that was that whatestimates were produced weren't relevantenough from the view of the policymaker, sowe went in another direction. And I think it

sounds like what we're saying is that we needto find that middle ground and perhaps getback into the system more of the strategic,long-term, basic research that Sherman Kentadvocated for so many years.

* * *

I have yet to find the question that apolicymaker would like answered that oneanalyst can answer. Either it's too general, orit's too detailed. We need to rethink the way weput our analysts together, so that they canbetter respond to the information and thequestions they are receiving.

* * *

But there's a difference, which is that theIntelligence Community is supposed tocontinue to criticize after the decision has beenmade, which is more difficult to do. [That] putsthem in a very different position from themilitary. They're not supposed to salute andsay, "Yes, sir." They're supposed to tell him hedoesn't have any clothes on every day.

* * *

Part of our job is to tell the emperor he has noclothes. But, once the emperor's in the stewpot, you're not doing him any good saying overand over again, "You've got no clothes." . . .Start thinking about what we need to be payingattention to [in order] to achieve whatever thegoal was at the end of the day. And if you're anintelligence analyst and you can't get off theissue of whether this is a good or bad decision,it's really, really tough. At that point, seniorofficials in the Agency have to kind of jerk thewhole system and get them to focus on theother thing. And then, what does that look liketo the analyst down below? That'spoliticization.

* * *

We need torethink the waywe put ouranalyststogether, so thatthey can betterrespond to theinformation andthe questionsthey arereceiving.

Once theemperor's in thestew pot, you'renot doing him anygood saying overand over again,"You've got noclothes."

10

There's been no fundamental change in the DIpersonality. This contrariness is still there, and,I think, it's something we encourage. It doescome down more to core values. And you dohear in the Kent School and CIA University theimperative of speaking truth to power . . . .

* * *

I have found, at least, that the counter often, inthe covert area, was the DDI's willingness tostrike an independent stance, independent ofthe covert action, and assess it. . . . If you wantto have real problems, send the covert actionpeople off by themselves. We've had that, youknow. We had the Bay of Pigs, where therewas no DI involvement.

* * *

We're putting analysts much closer tooperators, both on covert action and operationsin an effort to get smarter operations andsmarter covert action . . . . The risk, of course,is that you begin to blur these lines. And are yousomehow co-opting the analytic function? Onereason why we've succeeded . . . in not doingthat is we've maintained the analytic unitsseparate from those units with analysts in themthat are doing the support to operations,support to covert action.

Challenges for the DCI

Presidents Clinton and Bush charged DCITenet with carrying out an active and visiblerole within the framework of the Middle Eastpeace process and the war on terror.Roundtable participants were divided as to theadvisability of DCIs assuming this kind ofresponsibility. Among the reservationsexpressed were that such involvement tookvaluable time from the DCI's statutory dutiesas manager of the Intelligence Community andthat it created the risk that the IntelligenceCommunity's objectivity on that issue wouldbe—or might appear to be—compromised.

In another area touching on DCIresponsibilities, there was an extendeddiscussion as to whether there were newrealities in the post-9/11 world that mightnecessitate a reordering of responsibilitieswithin the Intelligence Community. Should theFBI, for example, be divided into an entity withdomestic counterintelligence responsibilitiesand another dedicated to its traditional lawenforcement function? This question,although hardly new, has become more salientduring the past several years becausedistinctions between foreign and domesticintelligence have become increasingly blurredand previous conflicts among agencies,particularly the FBI and the CIA, over the usesof intelligence have become better known.

While recognizing the need to address theseproblems, several participants expressedconcern that attempted institutional fixes forIntelligence Community problems, such asreorganizations or creating new agencies,would simply further complicate an alreadycomplex picture and permit new, probablyunforeseen, dangers to arise. One speakermentioned that there had been reneweddiscussion on the advisability of creating theposition of Director of National Intelligence withgenuine authority over all the Communityagencies and the entire Community budget,but several other panelists argued the risks toanalytic objectivity and diversity of opinion if acentral authority were established.

Roundtable participants agreed that thelikelihood of another terrorist attack within theUnited States was substantial, especiallyduring the run-up to the national election inNovember 2004. Several foresaw that, in thatevent, the Intelligence Community mightbecome an easy target of political partisanslooking for a scapegoat. This would be all themore likely if a case could be made that asuccessful attack had been the result of anIntelligence Community failure to correctdefects that had been pointed out earlier. Theonly way to prevent such scapegoating, they

There's been nofundamentalchange in the DIpersonality. Thiscontrariness isstill there, and, Ithink, it'ssomething weencourage.

We're puttinganalysts muchcloser tooperators, bothon covert actionand operations inan effort to getsmarteroperations andsmarter covertaction.

11

agreed, would be for a senior administrationofficial to make clear to the Congress and thepublic that there are limits to what theIntelligence Community can do to foil terroristattacks.

Continuing the discussion of a possible secondforeign terrorist attack on the United States, aparticipant commented that the Patriot Act hadimproved the chances of heading off anattempt but expressed concern that the newauthorities granted by the act might tend toinfringe on civil liberties. A speaker suggestedthat not enough had yet been done by federalintelligence and law enforcement agencies tosupport "first providers" throughout thecountry. Another participant commented thatthis could become one of the responsibilities ofthe new Terrorism Threat Information Center(TTIC) but that questions still remained as towhether the center would focus on analysis ordissemination.

Discussion Excerpts

If the Director of Central Intelligence isresponsible for implementing some aspect ofthe policy, whether he's implementing it overtlyor covertly doesn't really make too muchdifference; he becomes almost axiomatically aproponent of the policy he's implementing . . . .Assuming he does believe in what he's doing,it becomes much less likely that theIntelligence Community will provide a unifiedproduct saying that whatever we're doing is abad idea . . . . You just have to accept that, andit's a question of educating the consumersabout what to expect from a bureaucraticarrangement that has those elements to it. Thisgets to a broader question, which is, is theIntelligence Community supposed to besupporting policy, or is it essentially anadversarial function in which it is supposed tobe finding weaknesses and vulnerabilities?

* * *

I'd like us to clarify if the Agency was looking forcovert action jobs to do. I would say that theyshowed about the same degree of enthusiasmas the Joint Chiefs did for any military action,certainly in the last four years of the Clintonadministration when I was directly involved.And you probably know that the Chiefs' answerto any military proposal from the StateDepartment was, "Four divisions, four hundreddays, four hundred billion dollars."

* * *

What we've got now is the pressure to beperfect, and what we've got in the Patriot Act[are] some tools that allow us to up ourchances of performing against that standard.But there's a trade-off, and . . . to the degreethat you move it [sic] in one direction, youimprove your chances of preventing another9/11 . . . at the risk of personal liberties andprivacy and potential for abuse. You move it[sic] in the other direction you improve thoseguarantees, but you open yourself here.

* * *

I think that, looking for institutional fixes, onehas to be rather careful and recognize that, infixing one problem, you're going to createanother problem. In other words, by shiftingorganizational boundaries, you'll better handlethis issue, but you'll create other things that'llfall between cracks that previously were quitewell handled, because you've simply movedthose boundaries. But there are always goingto be boundaries. If the problem that you'readdressing is sufficiently consequential, then itprobably does make sense. So, if you're tryingto prevent another 9/11-type catastrophe, thencreating a Department of Homeland Security. . . is a defensible response, even though itcreates a lot of other lacunae in whichproblems will develop, because there's nolonger somebody who's focusing most of hisattention on taking care of those problems.And, similarly, I'm sufficiently persuaded by thelogic of it to suggest that you need a domestic

Is the IntelligenceCommunitysupposed to besupportingpolicy, or is itessentially anadversarialfunction in whichit is supposed tobe findingweaknesses andvulnerabilities?

12

intelligence agency if your object is to preventattacks, not prosecute the perpetrators. If yourpriority attention is to that, the tension betweenthe two functions is sufficient to argue that thetwo shouldn't be in the same agency.

* * *

One of the things it seems to me is likely tohappen out of all this is a strengthening of theDCI authorities and perhaps a movementtoward . . . trying the DNI [Director of NationalIntelligence] kind of theory. Well, one of theserious questions is whether that addresses inany fundamental way the real problems thatwe've been describing. It's not obvious to methat it does.

* * *

One of the notions behind TTIC was to be asort of a mechanism to get information at thenational level down to first responders . . . . Andthere's a real debate within our own house overwhat the nature of that commission really is atthe end of the day: whether it's basically anintegrator and disseminator of information orwhether it's an originator of original sourcethings. It used to be that what was reallyimportant was providing and serving thepresident. Now it's clear that a great deal ofwhat the mission is is the dissemination ofactionable intelligence, not just to thePentagon, but down to the guy who's makingthe traffic stop in Winslow, Arizona. And we'renot trained, we're not equipped, we're not inhand for that mission.

* * *

We are now subject to terrorist attacks that aredesigned to disrupt and manipulate ourinternal political process. I believe there is areal risk that terrorist groups will try to exploit

our election season . . . . This is somethingwe've never faced before—an external non-state actor or set of actors trying to influenceour political processes.

* * *

I'm still convinced that the biggest set ofpotential failures doesn't have to do withoverlap; it has to do with underlap, that is,failures to meet the exigencies and demands ofthe intelligence requirements for the policyprocess and, quite simply, on the other side,failure to meet the exigencies and demands ofwhat's the best selection of policy options inthe national interest, broadly looked at, whichis constantly, obviously, under stress. At thepresidential level, its reelection on the onehand and, secondly, coming up with a policythat serves the national interest but also thatdoesn't hurt in being reelected . . . .

* * *

I would agree with you that the expectation haschanged since 9/11. And somehow we havebuilt up this myth that we know we have thebest intelligence in the world, but that we can'tknow everything. And I think at some pointthere is going to have to be some sort of, well,probably not a speech, but something to talk tothe American public, maybe in the politicalprocess, about what intelligence can do andwhat it can't do, to this point of leveling with theAmerican public.

A great deal ofthe mission isdissemination ofactionableintelligence, notjust to thePentagon, but tothe guy makingthe traffic stop inWinslow,Arizona. Andwe're not trained,we're notequipped, we'renot in hand forthat mission.

13

The Elusive Red Line

During the discussion of the more active andpublic role assigned to DCI Tenet byPresidents Clinton and Bush, a participantasked if, as a practical matter, this new taskingmeant that DCIs are no longer subject to the"red line," the customary demarcation betweenthe intelligence and policy functions. Althoughopinion on this issue was divided, noparticipant took the position that these newresponsibilities were necessarilyinappropriate. In fact, there was generalagreement that the idea of a red line issomewhat artificial, especially in the newcircumstances since 9/11, and that efforts toimpose rigid rules on the intelligence-policydialogue are likely to prove frustrating.

One roundtable participant favored a flexibleapproach that recognized that the activeinvolvement of intelligence in a "good" policyinitiative would improve the results. (Leftunsaid was the obverse of this proposition: thatthe results of a "bad" policy would presumablybe worsened by intelligence involvement.)Another participant commented that he foundrecently documented derelictions of bothpolicymakers and intelligence organizationsand their mutual failures to communicate farmore troubling than transgressions of a redline that, for him, was difficult to define. A thirdspeaker saw a loss of competition andtransparency as more dangerous for theIntelligence Community than red line issues.

One participant with both policy andintelligence experience provided an exampleof a kind of "reverse" red line. While he had noobjection to policymakers' "intelligentquestioning" of analytic judgments, he stronglycriticized efforts by policymakers to "bowl over"analysts by using data selectively.

Roundtable participants inclined more towardapproaching red line issues as a matter both ofprinciples and of checks and balances. Withrespect to the former, for example, severalparticipants noted that assisting a policymakerto realize a policy objective might well alsopromote the political objectives of the

policymaker or his political party. A strictinterpretation of the red line concept mightpreclude such assistance, the speakers noted,but they insisted that taking this political realityinto account did not necessarily mean that anintelligence officer's judgments would becompromised. With respect to checks andbalances, participants developed a substantiallist of actors (including Congress, the media,the public, and academics) and institutionalfactors (such as organizational structures andinternal bureaucratic disputes) that serve mostof the time to force policy and intelligence tohew to their accepted roles.

The discussion led to a consensus that theIntelligence Community must recognize the riskof staying close to the policy community but thatthe potential gains from keeping them inproximity provide ample justification for doingso. In the end, because of the ambiguities theyhad identified during the day, roundtableparticipants preferred to see the red line as amore neutral shade, such as gray, and as a linewithin the intelligence-policy relationship andnot between the two.

Discussion Excerpts

That, at least, raises for me the question ofwhether there is a red line if it comes to theDCI, or whether we're talking about exceptionsat the top and, then, everyone else has to workaccording to certain rules.

* * *

I think that intelligence officers too easily wantto blame the policymaker for their problems.I'm not saying the policymaking process iseven logical, let alone perfect, but I think thatmany of the problems that we face asintelligence officers we've got to deal with . . .ourselves and accept the reality that the policyworld is out there, and we have to interact with

There wasgeneralagreement thatthe idea of a redline is somewhatartificial, espe-cially in the newcircumstancessince 9/11.

I think thatintelligenceofficers too easilywant to blame thepolicymaker fortheir problems.

14

it . . . . We've got to solve many of our ownproblems, and they have nothing whatsoeverto do with red lines or with the policy process.

* * *

But the idea that intelligence can ignore thepolitical atmosphere in which it's beingdelivered is, again, a Panglossian affliction.

* * *

We were not the least bit reticent in saying,"You go this way, you're going to have a wreck."But it wasn't a case we were saying the policywas good or bad. That was supposed to comefrom somewhere else. And I think it's not amatter of a red line; it's just that there weredifferent jobs.

* * *

From the discussion, I get the impression thatthe failures of the intelligence side and thefailures on the policymaking side, and maybesome of their inability to communicate, areperhaps more serious than the transgressionsof a red line, which is hard for us to define at thepresent time.

* * *

If we ever take away the transparency andcompetition, then that will be a lot moredangerous than any red line issue. I don'tmean transparency giving away secrets; Imean everybody's opinion . . . any wackowants to come in is fine as long as you've gotto go up to the plate just like the rest of us.

* * *

You could say that, in administrations that tendto look at cooperation and hearing all views,the red line problem is real. In administrationsthat don't care to hear all views and [tend] not

[to] listen to a different set of approaches andattitudes, it isn't the red line problem that'smost important.

* * *

In President Clinton's administration, therewas an effort to reach out and coordinate andmake sure everybody's views were heard, [toreach out] to the academic community to makesure that competing views and assessmentswere brought into the equation as [the]decisionmaking process was moving on . . . .there was a concerted effort at that kind ofconsensus building.

* * *

The other problem that we have . . . is theproblem of the Intelligence Community beingcoopted by the policy community and thatdistorting the process to the point that it yieldscounterproductive results. And perhaps herethe . . . answer is a more sophisticatedunderstanding on the part of the Congress andthe American people of how these institutionsoperate and what their limitations are and therecognition that the Intelligence Communitywill occasionally be coopted by the policycommunity, as a result of which its findings andconclusions will be less reliable and simplyaccept that as a consequence of the necessityof the two to interrelate on a fairly intensebasis.

* * *

In the end, the person you elect as president isgoing to have to determine a lot of this. And,like a lot of other things in this country, there'sno organization chart. There may be a set ofrules in the Constitution and elsewhere, but Idon't think there's an easy way to say, "Here'sthe red line. Thou shalt not cross."

* * *

I get theimpression thatthe failures of theintelligence sideand the failureson the policy-making side, andmaybe some oftheir inability tocommunicate,are perhaps moreserious than thetransgressions ofa red line.

15

First Speaker: That red line is certainly not abright one, and it might not even, who knows, itmay not even be red. You know, it kind of maybe some other shade.

Second Speaker: Yellow or amber.

Third Speaker: Probably gray.

Voice from the audience: Probably gray, right.

* * *

If we were still living in the world of September10, I think there would have been quite aspirited debate on the nature of the red line,where the red line is today. But, as we foundout quickly in the afternoon, in many ways,we've moved beyond that. And it may comeback to the fundamental issue that we're atwar.

17

Key Questions

What do policymakers want from intelligence?Consider the full spectrum covering rawintelligence/finished intelligence includinganalysis/multiple policy options andimplications/policy advice/policy advocacy?

What factors determine the extent to whichpolicymakers expect a greater/lesser role forintelligence in the policy process? Forexample, is the expectation dependent onwhether the policy issue concerns a deniedarea (closed society or no regular diplomaticcontact)? How about issues that involveextensive covert action (Afghanistan)?

How do policymakers deal with unwantedpolicy advice from the IntelligenceCommunity? Ignore it? Set up a separateintelligence shop? Stop inviting intelligenceofficers to policy deliberations? What happenswhen intelligence officers refuse to give policyadvice?

How do senior intelligence officers see theirrole along the spectrum from informationprovider to adviser to advocate in policyformulation and implementation?

How does this role change when covert actiontransforms the DCI into a force provider/commander?

What criteria should govern future decisionson crossing the red line between intelligenceand policy? Is there now or should there evenbe a red line when homeland security isinvolved?

Should a new "red line" be established?Where should it be? Should intelligenceofficers assume a primary role for formulatingpolicy? Should they make policy suggestions?Should there be a distinction between the two?

How does a policy role for intelligence officersaffect their objectivity?